The Problem
Astronomical bodies can be hard to detect with our naked eye. Even with our most advanced tools, we are unable to recognize and detect certain bodies – and more specifically, Category C comets. Thanks to the rise of Artificial Intelligence (AI), we can now use Convolutional Neural Networks (CNNs) to detect these comets.
Detecting Comets with Convolutional Neural Networks (CNNs)
Convolutional Neural Networks (CNNs) are a class of deep learning algorithms that have proven to be highly effective in image recognition tasks. By training a CNN on a large dataset of astronomical images, it is possible to identify patterns and features that indicate the presence of comets. I focused on using CNNs to detect comets in astronomical images, particularly those that are faint or partially obscured, using publicly available datasets from NASA.
Training the CNN
I trained the CNN on a dataset of astronomical images that contained both positive and negative examples of comets. Each set contained 5 images, which include the faint comets at times. The CNN was able to learn and recognize the pixel-sized differences on the images, and to distinguish the comets from other objects in the images. By using a large dataset and training the CNN on a powerful GPU, I was able to achieve high accuracy in detecting comets in the images.
Technical Details
- Language: Python
- Libraries: TensorFlow, AstroML and Astropy (for astronomical data analysis)
Results
I compared three various CNNs with different loss functions: Mean Squared Error (MSE), Mean Absolute Percentage Error (MAPE) and Mean Absolute Error (MAE). The CNN with the MAE loss function performed the best, achieving an accuracy of 75% in detecting comets in the images.